first

app.py

@@ -0,0 +1,77 @@
+import gradio as gr
+import torch
+from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
+from PIL import Image
+from byaldi import RAGMultiModalModel
+from qwen_vl_utils import process_vision_info
+
+# Model and processor names
+RAG_MODEL = "vidore/colpali"
+QWN_MODEL = "Qwen/Qwen2-VL-7B-Instruct"
+
+def load_models():
+    RAG = RAGMultiModalModel.from_pretrained(RAG_MODEL)
+    
+    model = Qwen2VLForConditionalGeneration.from_pretrained(
+        QWN_MODEL,
+        torch_dtype=torch.bfloat16,
+        attn_implementation="flash_attention_2",
+        device_map="auto",
+        trust_remote_code=True
+    ).eval()
+    
+    processor = AutoProcessor.from_pretrained(QWN_MODEL, trust_remote_code=True)
+    
+    return RAG, model, processor
+
+RAG, model, processor = load_models()
+
+def document_rag(image, text_query):
+    messages = [
+        {
+            "role": "user",
+            "content": [
+                {
+                    "type": "image",
+                    "image": image,
+                },
+                {"type": "text", "text": text_query},
+            ],
+        }
+    ]
+    text = processor.apply_chat_template(
+        messages, tokenize=False, add_generation_prompt=True
+    )
+    image_inputs, video_inputs = process_vision_info(messages)
+    inputs = processor(
+        text=[text],
+        images=image_inputs,
+        videos=video_inputs,
+        padding=True,
+        return_tensors="pt",
+    )
+    inputs = inputs.to(model.device)
+    generated_ids = model.generate(**inputs, max_new_tokens=50)
+    generated_ids_trimmed = [
+        out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
+    ]
+    output_text = processor.batch_decode(
+        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
+    )
+    return output_text[0]
+
+# Define the Gradio interface
+iface = gr.Interface(
+    fn=document_rag,
+    inputs=[
+        gr.Image(type="pil", label="Upload an image"),
+        gr.Textbox(label="Enter your text query")
+    ],
+    outputs=gr.Textbox(label="Result"),
+    title="Document Processor",
+    description="Upload an image and enter a text query to process the document.",
+)
+
+# Launch the app
+if __name__ == "__main__":
+    iface.launch()

requirements.txt

@@ -0,0 +1,11 @@
+streamlit
+torch
+torchvision
+torchaudio
+torchao
+git+https://github.com/huggingface/transformers.git
+diffusers
+Pillow
+byaldi
+qwen_vl_utils
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.9.post1/flash_attn-2.5.9.post1+cu118torch1.12cxx11abiFALSE-cp310-cp310-linux_x86_64.whl